diff --git a/examples/peripherals/dedicated_gpio/soft_uart/components/soft_uart/include/soft_uart.h b/examples/peripherals/dedicated_gpio/soft_uart/components/soft_uart/include/soft_uart.h
index d35dc6362f..8d3e9f1601 100644
--- a/examples/peripherals/dedicated_gpio/soft_uart/components/soft_uart/include/soft_uart.h
+++ b/examples/peripherals/dedicated_gpio/soft_uart/components/soft_uart/include/soft_uart.h
@@ -65,6 +65,10 @@ esp_err_t soft_uart_del(soft_uart_port_t port);
 /**
  * @brief Send the given bytes on the software UART port.
  *
+ * @note Since toggling fast GPIOs for the first time may be slow (~1us), the first byte may be corrupted.
+ *       If you are seeing this issue, prepend a dummy byte to the buffer to send when calling this
+ *       function for the first time.
+ *
  * @param port Software UART port to send data on.
  * @param write_buffer Buffer containing the bytes to send on the buffer. Must not be NULL.
  * @param write_size Size of the write buffer. Must not be 0.
diff --git a/examples/peripherals/dedicated_gpio/soft_uart/components/soft_uart/riscv/soft_uart.S b/examples/peripherals/dedicated_gpio/soft_uart/components/soft_uart/riscv/soft_uart.S
index 4440886f7f..afb0f56be8 100644
--- a/examples/peripherals/dedicated_gpio/soft_uart/components/soft_uart/riscv/soft_uart.S
+++ b/examples/peripherals/dedicated_gpio/soft_uart/components/soft_uart/riscv/soft_uart.S
@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: 2010-2022 Espressif Systems (Shanghai) CO LTD
+ * SPDX-FileCopyrightText: 2010-2024 Espressif Systems (Shanghai) CO LTD
  *
  * SPDX-License-Identifier: CC0-1.0
  */
@@ -33,11 +33,6 @@ emulate_uart_send:
     sll a2, t0, a2
     /* Save return address in a4 */
     mv a4, ra
-    /* As UART is very time sensitive, we want each bit sent to be very precise in terms of duration.
-     * The first toggle of the fast GPIO register may be slow, ~1us, so let's send a dummy byte here
-     * before the actual UART emulation start, else the first byte sent would be corrupted.*/
-    li t0, 0
-    call uart_send_byte
     /* Reading the characters 4 by 4 would be much faster, but in our case, we don't need
      * the process to be fast as the bottleneck is the UART speed */
 uart_read_next:
diff --git a/examples/peripherals/dedicated_gpio/soft_uart/main/soft_uart_main.c b/examples/peripherals/dedicated_gpio/soft_uart/main/soft_uart_main.c
index e21018dcb9..860ee07b75 100644
--- a/examples/peripherals/dedicated_gpio/soft_uart/main/soft_uart_main.c
+++ b/examples/peripherals/dedicated_gpio/soft_uart/main/soft_uart_main.c
@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: 2010-2022 Espressif Systems (Shanghai) CO LTD
+ * SPDX-FileCopyrightText: 2010-2024 Espressif Systems (Shanghai) CO LTD
  *
  * SPDX-License-Identifier: CC0-1.0
  */
@@ -18,6 +18,7 @@ void app_main(void)
     esp_err_t ret = ESP_OK;
     uint8_t read_buffer[READ_COUNT] = { 0 };
     const uint8_t write_buffer[] = "Hello, world! This is a message.\r\n";
+    uint8_t dummy = 0;
     soft_uart_port_t port = NULL;
     soft_uart_config_t config = {
         .tx_pin = CONFIG_EMULATE_UART_GPIO_TX,
@@ -29,6 +30,10 @@ void app_main(void)
     ret = soft_uart_new(&config, &port);
     ESP_GOTO_ON_ERROR(ret, error, EXAMPLE_TAG, "Error configuring software UART");
 
+    /* Send a dummy byte as the software UART driver may corrupt the first byte */
+    ret = soft_uart_send(port, &dummy, sizeof(uint8_t));
+    ESP_GOTO_ON_ERROR(ret, error, EXAMPLE_TAG, "Error writing a dummy byte");
+
     /* Write few bytes to the UART */
     ret = soft_uart_send(port, write_buffer, sizeof(write_buffer));
     ESP_GOTO_ON_ERROR(ret, error, EXAMPLE_TAG, "Error writing to the software UART");